onezion-workspace-cleanup

---

name: onezion-workspace-cleanup description: Scan and clean up bloated WorkBuddy workspaces. Uses dust/dua CLI for fast disk analysis. Identifies large files (videos, node_modules, .venv, build artifacts), generates a report, and optionally removes them with user confirmation. Trigger on: "清理工作区", "磁盘不够了", "workspace cleanup", "free up space", "工作区膨胀", "node_modules 太大". allowed-tools: Bash, Read, Write, AskUserQuestion agent_created: true

Workspace Cleanup Skill

Scan WorkBuddy workspaces for bloated directories and large files, generate a report, and optionally clean them up.

CLI Tools (pre-installed)

Tool	Purpose	Install
dust	Tree map view of disk usage, top-N biggest items	brew install dust
dua	Fast parallel disk usage analyzer, supports batch delete	brew install dua-cli
du	macOS built-in, fallback	built-in
df	Filesystem-level space check	built-in

Step 1: Disk overview

# Overall disk status
df -h /

# WorkBuddy total + top 20 biggest workspaces
du -sh ~/WorkBuddy/ 2>/dev/null
du -sh ~/WorkBuddy/*/ 2>/dev/null | sort -rh | head -20

Step 2: Visual scan with dust

# Top 20 biggest items in WorkBuddy (tree view)
dust -n 20 ~/WorkBuddy/

# Top 15 biggest items in a specific workspace
dust -n 15 ~/WorkBuddy/<workspace-id>/

Step 3: Deep scan — find the usual suspects

# Find large files (>50MB) — videos, binaries, archives
find ~/WorkBuddy -type f -size +50M -exec ls -lh {} \; 2>/dev/null | sort -k5 -rh

# Find all node_modules directories with sizes
find ~/WorkBuddy -name "node_modules" -type d -exec du -sh {} \; 2>/dev/null | sort -rh

# Find Python venv directories
find ~/WorkBuddy -name ".venv" -type d -exec du -sh {} \; 2>/dev/null | sort -rh

# Find build artifacts
find ~/WorkBuddy \( -name ".next" -o -name "dist" -o -name "build" -o -name "__pycache__" -o -name ".pytest_cache" \) -type d -exec du -sh {} \; 2>/dev/null | sort -rh

# Find video files specifically (the #1 space killer)
find ~/WorkBuddy \( -name "*.mov" -o -name "*.mp4" -o -name "*.avi" -o -name "*.mkv" \) -exec ls -lh {} \; 2>/dev/null | sort -k5 -rh

Step 4: Generate report

Present findings to the user in a clear table:

• File path (relative to ~/WorkBuddy)
• Size
• Type (video / node_modules / venv / build artifact / other)
• Recommended action (delete / move to external SSD / keep)

Step 5: Confirm before any cleanup

NEVER delete without user confirmation. Use AskUserQuestion to let the user select which items to clean.

For each confirmed item:

• Videos (.mov, .mp4): Move to trash via osascript -e 'tell application "Finder" to delete POSIX file "PATH"'
• node_modules: rm -rf (can always be regenerated with npm install or pnpm install)
• .venv: rm -rf (can be regenerated with python3 -m venv .venv)
• Build artifacts (.next, dist, build, pycache): rm -rf (regenerated on next build)
• Entire completed workspaces: tar czf to external SSD first, then trash

Step 6: Post-cleanup verification

# Verify space recovered
df -h /
du -sh ~/WorkBuddy/
dust -n 10 ~/WorkBuddy/

Automation suggestion

If the user wants ongoing management, suggest creating an automation that runs this scan weekly:

• Schedule: weekly on Sunday
• Action: scan and report only (never auto-delete)

Key rules

1. Always scan first, act second. Never skip the report.
2. Always confirm before deletion. List every file path.
3. Prefer trash over rm. Use osascript Finder delete for user files.
4. node_modules and .venv are safe to delete — they're reproducible.
5. Videos are the #1 space killer. Always check for .mov and .mp4 files first.
6. Never touch project source code, configs, or .git directories without explicit approval.
7. Use dust for quick visual scans, dua for detailed per-directory analysis, du as fallback.